ERP X Is Over. But Its Two-Year Test Proved a Point

The cameras worked, drivers liked the app, and nobody needed another box in the car. So why is Singapore not rolling it out?

Source: erpx.open.gov.sg

Singapore has charged drivers for using congested roads since long before apps and smartphones. The Electronic Road Pricing system, or ERP, arrived in 1998. Drive under a gantry at the wrong time, the in-vehicle unit beeps, and money comes off your CashCard.

That system is reaching the end of its life. Its replacement, ERP 2.0, was awarded in 2016 and uses satellite positioning instead of relying on physical gantries. It also comes with a new on-board unit (OBU) fitted inside every vehicle.

The rollout has not exactly been popular. Drivers questioned the three-piece setup, the extra dashboard screen and why a modern road-pricing system still needed so much dedicated hardware when most people already carry a connected computer in their pocket.

Inside the Government, a small team was asking the same thing.

Where ERP X came from

ERP X began in January 2024 during Hack for Public Good, an annual hackathon run by Open Government Products (OGP). OGP is an experimental team within the Singapore Government that builds and tests public-service technology.

Several people on the team were drivers or riders themselves. Instead of starting with a long specification, they tried to answer one practical question:

Could Singapore charge for road use with cameras and a phone app, without installing another device in every car?

Source: erpx.open.gov.sg

Their first version used an ordinary webcam bought from Sim Lim Square. The camera read a vehicle’s number plate using automatic number plate recognition, or ANPR. The driver registered the vehicle and a payment method in an app. When the plate was detected at a charging point, the transaction appeared on the phone.

No gantry needed to communicate with the car. No CashCard to top up. No additional box on the dashboard.

Within a month, the team had a working prototype. It later became a live experiment developed with the Land Transport Authority (LTA), first on ordinary roads around Bras Basah and eventually at active ERP locations on the expressways.

Two years later, ERP X had grown to 63 cameras covering all 22 active ERP charging locations, with 403 motorists testing the app.

The results were better than many people would have expected. The cameras read Singapore number plates correctly 97.16% of the time, while testers rated the overall experience 8.58 out of 10.

Then OGP ended the project.

The cameras are now switched off, and the Government says it has no plans to roll ERP X out. It was an experiment, not a late attempt to cancel ERP 2.0.

Your OBU is not going away. But ERP X has left Singapore with real evidence that road pricing can work another way.

The cameras did better than you might expect

The first prototype was properly scrappy: one webcam, a tripod, a laptop and team members driving their own vehicles along an empty road.

Cars and trucks were recognised 95% to 100% of the time, but motorcycles managed only around 85%. That gave the team something concrete to improve before moving on to public roads.

The next phase put 26 cameras around Bras Basah. Single lanes became dual lanes, traffic got heavier, and average accuracy improved from 95.4% to 97.3%.

From October 2025 to March 2026, the test expanded again. Sixty-three cameras covered all 22 active ERP sites, including expressways. Now the system had to deal with wider roads, faster vehicles, rain and darkness.

A single roadside camera could handle one or two lanes on a normal road without much trouble. Expressways were harder. Camera position mattered more, and wet or dark conditions hurt accuracy at higher speeds.

Source: erpx.open.gov.sg

The final result across all conditions was 97.16% for Singapore-registered vehicles. Good? Definitely. Good enough to charge every vehicle automatically with no backup process? Not quite.

At Singapore’s traffic volumes, even a small error rate can produce plenty of wrong or missed charges. A live system would still need to handle foreign vehicles, cloned or dirty plates, disputes and the odd camera having a bad day.

About that $2.7 million figure

OGP estimates that deploying ERP X at today’s 22 active sites would cost about $2.7 million upfront and $2.2 million a year to operate.

That is roughly $100,000 per site each year. Depending on the location, setup would cost between about $38,000 and $350,000. The estimate is based on what the pilot actually spent, with conservative figures used where the team could not pin down an exact amount.

You will probably see that compared with ERP 2.0’s reported $556 million contract. It makes for a great headline, but the comparison is not completely fair.

ERP 2.0 is an islandwide platform with hardware going into vehicles and other functions built around it. The ERP X estimate covers cameras at the 22 charging locations operating today. OGP itself says the pilot figures are not a guarantee of what a national system would cost.

Still, $2.7 million is enough to show that testing a different idea does not have to begin with another massive commitment. The Government now has actual numbers instead of a slide deck full of assumptions.

Drivers liked the boring part: it just worked

Over two years, 403 motorists joined the trial. A hundred of them completed the final survey, giving the experience an average score of 8.58 out of 10.

Why did they like it? Probably because there was very little to do.

No CashCard to top up. No extra screen competing with the phone already mounted on your dashboard. You could link multiple vehicles to one account and see the transactions in the app.

The testers also seemed to enjoy being included. Of the people who answered the final survey, 97% said taking part had been worth their time even though the project might never launch. Another 96% wanted to join similar government experiments in future.

Take those percentages with some salt. These were volunteers who chose to join an experimental government project, and only 100 of the 403 testers answered the final survey. They are not a stand-in for every driver in Singapore.

The trial shows that its users liked the app. It does not tell us how the wider public would feel about more cameras tracking vehicle movements, how long that data should be kept, or what happens when the system gets a plate wrong. Any serious rollout would have to answer those questions properly.

If it worked, why stop?

The blunt answer is that ERP 2.0 is already the system Singapore has committed to. ERP X was never promised a route into production, and OGP says the Government has no plans to roll out camera-based road pricing.

That does not make the last two years pointless.

Source: erpx.open.gov.sg

The team found its problems while they were still cheap problems: motorcycles were harder to recognise, expressways needed different camera layouts, and rain mattered more at speed. Drivers tried the app while the engineers could still change it.

Compare that with discovering the same issues after a specification has been frozen, a tender awarded and thousands of installations scheduled.

OGP’s real argument is about how government should build large technology projects: make a rough version, put it in the real world, let citizens use it, and learn what breaks before betting big.

ERP X will not replace ERP 2.0. But if Singapore revisits road pricing years from now, nobody has to guess whether a camera-based approach is possible. We have already tried it.

What this means for you

Right now, nothing changes. Keep the OBU appointment. Camera charging is not coming.

What has changed is what we know. Roadside cameras can recognise Singapore vehicles accurately enough to make road charging a serious option. The system can be built at a measurable cost, and the motorists who tested it generally preferred how little effort it required.

Whether Singapore should ever use it is a different question. Accuracy would need to improve or be backed by a fair appeals process. Privacy and journey data would need clear limits. And the cost would need to be tested beyond 22 existing sites.

ERP X did not give us the next ERP system. It gave us something more modest, and still useful: proof that there is another way to build one.

Read OGP’s full ERP X report:

https://erpx.open.gov.sg/

====================================

Join our community of over 20,000+ members today on Telegram! 🏎️ sgcarmeets.t.me 💨 🚗 sgcarnews.t.me 💨 💬 Chat with us: sgcmchat.t.me

Learn more about us at **autosave.club**